Differentially Private Feature Selection via Stability Arguments, and the Robustness of the Lasso thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Differentially Private Feature Selection via Stability Arguments, and the Robustness of the Lasso

Published on Aug 09, 20133787 Views

We design differentially private algorithms for statistical model selection. Given a data set and a large, discrete collection of “models”, each of which is a family of probability distributions, the

Related categories

Chapter list

Private Model Selection via Stability Arguments and the Robustness of Lasso00:00
What Can We Learn Privately? - 100:06
What Can We Learn Privately? - 201:17
Model Selection01:42
Running Example: Sparse Linear Regression - 102:08
Running Example: Sparse Linear Regression - 203:08
This Paper03:39
Why is Privacy a Concern in Model Selection? - 104:01
Why is Privacy a Concern in Model Selection? - 204:18
Why is Privacy a Concern in Model Selection? - 304:45
Di erential Privacy04:57
Prior Works on Learning and Privacy06:35
Our Contributions - 107:37
Our Results for Sparse Linear Regression09:13
Our Contributions - 211:10
Perturbation Stability based Model Selection - 111:14
Perturbation Stability based Model Selection - 212:32
Perturbation Stability based Model Selection - 313:08
Perturbation Stability based Model Selection - 413:35
Our Contributions - 314:19
Perturbation Stability of LASSO - 114:31
Perturbation Stability of LASSO - 215:02
Stability Test for LASSO 15:35
Geometry of LASSO - 116:41
Geometry of LASSO - 216:56
Making the Test Private (Simplifi ed)17:12
Putting the Pieces Together17:36
Revisit: Our Contributions18:13
Future Work18:36